﻿ 基于光纤光栅传感阵列和BP神经网络的悬臂梁裂缝损伤识别 Crack Damage Recognition of Cantilever Beam Based on Fiber Bragg Grating Sensing Array and BP Neural Network Algorithm

Journal of Sensor Technology and Application
Vol.05 No.01(2017), Article ID:19481,6 pages
10.12677/JSTA.2017.51003

Crack Damage Recognition of Cantilever Beam Based on Fiber Bragg Grating Sensing Array and BP Neural Network Algorithm

Pei Luo

The Optical Fiber Sensing Technology National Engineering Laboratory, Wuhan University of Technology, Wuhan Hubei

Received: Dec. 16th, 2016; accepted: Jan. 6th, 2017; published: Jan. 9th, 2017

ABSTRACT

The sensing theory of fiber Bragg grating strain sensing array and the BP neural network algorithm have been introduced in this paper. The strain changes of fiber Bragg grating sensing array measuring metal cantilever beam under the status of damaged and undamaged have been studied. Taking changes of strain as input data of neural network, the damage recognition of metal cantilever beam has been carried on using BP neural network algorithm through training and studying the neural network. The realistic output of Bp network is very near to the anticipant output. It indicates that the damage recognition of metal cantilever beam can be realized using fiber Bragg grating sensing array and BP neural network algorithm.

Keywords:Fiber Bragg Grating Sensing Array, Cantilever Beam, Cracks, Damage Recognition, BP Neural Network

1. 引言

2. 光纤光栅的传感原理

(1)

3. BP神经网络算法

BP算法 [9] [10] 是目前最常用的算法之一。它是1985年提出的一种监控式学习方法，BP神经网络的学习采用误差反向传播算法(Back Propagation，简称BP算法)，其主要思想是把整个学习过程分为四个部分：一是输入模式对从输入层经隐含层传向输出层的“模式顺传播”过程；二是网络的希望输出()与实际输出()之差的误差信号()由输出层经隐含层向输入层逐层修正连接权的“误差逆传播”过程；三是由“模式顺传播”和“模式逆传播”的反复交替进行的网络“记忆训练”过程；四是网络趋向收敛即网络的全局误差(E)趋向极小值的“学习收敛”过程，当网络全局误差E小于预先设定的极小值，网络收敛。只有BP神经网络在学习训练后实现收敛，才能说明所建立的神经网络模型是有效的。

4. 金属悬臂梁裂缝损伤数据的实验测试和数据分析

4.1. 金属悬臂梁损伤检测的实验装置的搭建

4.2. 实验数据的采集

4.3. 金属悬臂梁损伤识别的BP神经网络的实现

Figure 1. Sensing schematic of distributed fiber Bragg grating

Figure 2. Experiment chart of fiber Bragg grating sensing array

5. 结论

Table 1. Normalized data of fiber Bragg grating sensing array detecting strain of cantilever beam under different damage status

Table 2. Comparison of anticipant output and realistic output

Crack Damage Recognition of Cantilever Beam Based on Fiber Bragg Grating Sensing Array and BP Neural Network Algorithm[J]. 传感器技术与应用, 2017, 05(01): 16-21. http://dx.doi.org/10.12677/JSTA.2017.51003

1. 1. 宋启根. 带裂缝梁和板的应力分析[J]. 上海力学, 1985(3): 17-26.

2. 2. 裴强, 丁英哲. 低温下钢结构裂缝损伤识别方法(II)[J]. 低温建筑技术, 2005(4): 100-101.

3. 3. 王术新, 姜哲. 裂缝悬臂梁的振动特性及其裂缝参数识别[J]. 振动与冲击, 2003, 22(3): 83-87.

4. 4. 李晓飞, 余音. 含横向裂纹悬臂梁的损伤检测[J]. 上海交通大学学报, 2010, 44(6): 735-739.

5. 5. 沈亚鹏, 唐照千. 裂纹对悬臂梁板振动频谱的影响[J]. 固体力学学报, 1982(2): 247-251.

6. 6. 张炜, 毛崎波, 聂彦平. 含任意数目裂纹梁的振动分析[J]. 机械设计与制造, 2012(10): 228-230.

7. 7. 詹亚歌, 吴华, 裴金诚, 杨熙春, 薛绍林. 高精度准分布式光纤光栅传感系统的研究[J]. 光电子•激光, 2008, 19(6): 758-762.

8. 8. 田石柱, 曹长城, 王大鹏. 光纤光栅传感器监测混凝土简支梁裂缝的实验研究[J]. 中国激光, 2013, 40(1): 1-5.

9. 9. 禹建丽, 卞帅. 基于BP神经网络的变压器故障诊断模型[J]. 系统仿真学报, 2014, 26(6): 1343-1349.

10. 10. 唐立力, 吕福起. 基于遗传算法的BP神经网络滚动轴承故障诊断[J]. 机械设计与制造工程, 2015, 44(3): 65-68.